More managers are rehearsing with AI ahead of sensitive meetings such as delivering poor performance notices, excluding employees from pay rises or discussing dismissals. They are not delegating the actual conversation to AI, but using it to check tone and how messages are delivered in advance.

CNBC reported on Sept. 12 that AI coaching tools range from text chatbots to voice-based roleplay. Managers can repeatedly practise the same message by assuming an employee who gets angry or reacts emotionally, and receive AI feedback on tone, brevity and phrasing. Some services also reflect data on an employee's traits to set up scenarios in more detail.

A lack of preparation by managers is also driving AI use. The Predictive Index surveyed 399 managers and 208 CEOs and corporate leaders in July and found 74 percent of executives believed managers handle difficult employee meetings with strong confidence. But 42 percent of managers said, "I know what I should say, but it is difficult how to say it," and only 20 percent said they would prepare alone without any framework or coaching.

A key advantage of AI is repeated practice. Managers can recreate the same situation multiple times regardless of time and place, and spot overly blunt expressions, repetitive speech habits and unnecessary filler sounds. Experts view preparation as more important than spontaneity in higher-risk conversations, and say AI can serve as a practice partner to check messages before an actual meeting.

Privacy risks are also growing. In the survey, 72 percent of managers said open AI is useful for preparing employee-related conversations, and 44 percent said they have entered employees' names or performance information. Only 45 percent of organisations had an official AI use policy. Experts advised, "Do not enter identifiable information such as employee names and specific performance materials into open chatbots, and check data retention and training policies of company-approved systems."

AI is unlikely to fully replace human coaches. In particular, for conversations with heavy emotional weight such as dismissals, it is viewed as preferable to recheck messages and legal risks with a senior manager or the human resources team even after AI practice.